Research map / Image generation / Editing and style transfer
SSCGAN: Facial Attribute Editing via Style Skip Connections
Wenqing Chu, Ying Tai, Chengjie Wang, Jilin Li, Feiyue Huang, Rongrong Ji
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add style skip connections to spatial-only model (ablation) | SSCGAN · CelebA test | average attribute generation accuracy (%): 82.46 -> 94.21 | Improved |
| add spatial information transfer to style-only model (ablation) | SSCGAN · CelebA test | average attribute generation accuracy (%): 92.26 -> 94.21 | Improved |
| add style skip connections instead of spatial only (ablation) | SSCGAN · CelebA test | average attribute generation accuracy (%): 82.46 -> 92.26 | Improved |
| predict style from encoder style plus attribute difference with residual learning, not attribute difference alone (ablation) | SSCGAN · CelebA test | average attribute generation accuracy (%): 90.03 -> 94.21 | Improved |
| use style features from every block rather than only the last encoder stage (ablation) | SSCGAN-mm vs SSCGAN-lm · CelebA test | average attribute generation accuracy (%): 87.08 -> 87.60 | Improved |
| use both mean and variance instead of mean only (ablation) | SSCGAN-lmv vs SSCGAN-lm · CelebA test | average attribute generation accuracy (%): 87.08 -> 88.07 | Improved |
| add residual learning to per-block style embedding (ablation) | SSCGAN-mm · CelebA test | average attribute generation accuracy (%): 87.60 -> 94.21 | Improved |
| use multiple skip connections instead of one at 16x16 (ablation) | SSCGAN-16 · CelebA test | average attribute generation accuracy (%): 93.63 -> 94.21 | Improved |
| use multiple skip connections instead of one at 128x128 (ablation) | SSCGAN-128 · CelebA test | average attribute generation accuracy (%): 77.79 -> 94.21 | Improved |
| use multiple skip connections instead of one at 8x8 (ablation) | SSCGAN-8 · CelebA test | average attribute generation accuracy (%): 88.28 -> 94.21 | Improved |
| use SSCGAN instead of STGAN | facial attribute editing · CelebA test | average attribute generation accuracy (%): 83.21 -> 94.21 | Improved |
| use SSCGAN instead of STGAN | facial attribute editing · CelebA test | FID: 4.78 -> 4.69 | Improved |
Later papers that compared with this method
2 later accepted papers report SSC-GAN as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- SphericGAN: Semi-Supervised Hyper-Spherical Generative Adversarial Networks for Fine-Grained Image Synthesis (CVPR 2022 · better on 1 of 1 shared result)
- High-Fidelity and Arbitrary Face Editing (CVPR 2021 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- StarGAN — StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation (CVPR 2018) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- L2M-GAN: Learning To Manipulate Latent Space Semantics for Facial Attribute Editing (CVPR 2021 · similarity 0.90)
- STGAN: A Unified Selective Transfer Network for Arbitrary Image Attribute Editing (CVPR 2019 · similarity 0.90)
- Image-to-Image Translation With Disentangled Latent Vectors for Face Editing (TPAMI 2023 · similarity 0.89)
- SDGAN: Disentangling Semantic Manipulation for Facial Attribute Editing (AAAI 2024 · similarity 0.89)
- Efficient 3D-Aware Facial Image Editing via Attribute-Specific Prompt Learning (ECCV 2024 · similarity 0.88)
- Latent Transformations via NeuralODEs for GAN-Based Image Editing (ICCV 2021 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
